<vagrantc>are any variables exposed which point at LIBEXEC ? <vagrantc>would it be bad form to add libexecdir to PATH ? <bjc>that location is normally for things that are called by other executables <vagrantc>shoudl probably just ask on the list, still trying to figure out the "correct" thing for diffoscope <atka>vagrantc: did you say you had a honeycomb lx2 on order? after much deliberation I've decided to order one... <jab>has anyone tried to package sirmail for guix? It's written in racket... <jab>I will check the next time I have my guix system open... <atka>the_tubular: slower than I'd like, but I'd say yes, deployed a guix server today on the other side of the planet which is cool <atka>no, I just had a client put a guix iso on a drive, boot the installer, set root password, and start the ssh daemon and I took it from there <atka>and open the needed port <the_tubular>I think that would have been a good use case for guix deploy, but I've never used it so I don't know what I'm saying :P <wdkrnls>Dear Guix, I saw in the cookbook a reference to using env to create a shell script which didn't inherit from the parent shell environment. <wdkrnls>But I do not really understand how to use it. <atka>but its nice to be able to deploy a config with all the users and temp passwords, ssh keys, wireguard tunnels, networking all pre set up <wdkrnls>For example, I am trying to start emacs from a different profile, but I cannot get it to start when I don't inherit from the system environment. <wdkrnls>But when I try to just source that, it didn't work. <atka>the_tubular: yeah, I plan on playing with deploy but new I could ssh into the installer so I took the path of least resistance :) <atka>s/new/knew, been a long day :) <vagrantc>atka: yeah, ordered one, presuming all goes well should be here in a couple months... <atka>vagrantc: oh so the 8 weeks lead time is correct or no? <vagrantc>atka: ordered a week or so ago ... hopefully it's not worse than 8 weeks :) <the_tubular>Yeah, I still haven't implemented wireguard on my guix machine <atka>oh that's good to hear, are you planning on pairing it with a gpu for use as a workstation? I'm trying to find some ECC memory and use it as a server/nas as well as test out some networking ideas <atka>I plan on getting some of solid run's other networking boards as well so I can get away from mikrotik <atka>I do like mikrotik but building my own stuff is more fun <worstname>reading the section on `guix deploy', if i want to deploy multiple guix system machines, how do i deal with wanting different hostnames and usernames for the machines? generating those differences declaratively? <atka>worstname: also what about formatting and uuid for esp etc <atka>I probably missed it but don't you have to format and partition your drives first? ***dgcampea-2 is now known as dgcampea
<atka>dunno, but I doubt it would handle ecrypted btrfs, per user subvolumes etc <atka>I'm interested in bcachefs as well, but btrfs has been good to me <the_tubular>Mind sending me code snippet for setting up wireguard <atka>the_tubular: I just use raid1 or raid1cX and faster storage, ie all ssds instead of raid10 <atka>raid1c3, raid1c4 etc, extra copies so you can afford to lose more disks <the_tubular>So basically it's just replicated data on every disk ? <atka>raid1 is two copies of the data, c3 is three copies, c4 4 copies <atka>data and metadata that is <the_tubular>What do you get having 4 mirror in the same physical location ? <atka>just more redundancy, I don't use it but c3 is good <wdkrnls>How do I start a shell with all the system environment variables plus the environment variables of a profile? <wdkrnls>It seems like I cannot start a shell without it either inheriting nothing (env -i) or inheriting the variables for the default Guix profile. <notzmv>folks, I need to patch a package by overwriting some files with modded ones, and I found the builder file and saved it locally elsewhere so I could edit. it uses gnu-build to do the actual building. would I then want to write a gexp after gnu-build or could I just use copy-recursively somehow? <vivien>notzmv, hello, what most people would do here would be to define a package that inherits the base package and redefine the origin so as to include the patch. See 8.3 Defining Package Variants in the manual, and 8.2.2 origin Reference for how to change that (see patches and snippet) <notzmv>ah, I hadn't noticed snippet, nor that url-fetch works fine with file://. I think I have enough to make this work now. Thanks, vivien! <meso>hello, I am trying to setup a emacs daemon shepherd service and was wondering if anyone can check it out to see if i am doing it correctly. https://paste.debian.net/1238167/ the issue I am having is that the service never shows its running with herd status, and when I do start it, it runs like 6 times when it was already succesful the first time it was executed. <meso>i followed the services writeup setting the shepherd/init.scm file with individual services in shepherd/init.d/service.scm <gooksnoob21>i was trying to install guix but after running guix system init there was an error and the installation stopped <gooksnoob21>the only output for the error was "guix system: error: cannot close compressed log file (BZip2 error = -6)" <xelxebar>Huh. `herd rules udev` no longer works: herd: service 'udev' does not have an action 'rules' <xelxebar>Also, I'm getting an error during /etc population of a guix system reconfigure. <xelxebar>shepherd/service.scm:776:0: In procedure exec-command: <xelxebar>Unrecognized keyword: #:supplementary-groups <Ashy>is there a way to generate a manifest file from a `guix environment ...` command? <bjc>just copy the arguments into a manifest.scm <Ashy>yeah i just realized how simple the manifest format is haha <bjc>sometimes it feels like guix is a dense block of scheme code that has to be chipped away at to achieve something, but in this case it's incredibly obvious <Ashy>compared to nix it's an amazing upgrade in clarity <bjc>i haven't used nix, so i'll take your word for it <Ashy>is there a convention for naming the manifest files? <bjc>manifest.scm or guix.scm <bjc>guix will automatically use them with ‘guix shell’ if they're in your safe load path <bjc>(guix environment is going away eventually, btw, and guix shell is the replacement) <Ashy>hmm, i wonder which texlive-latex-* packages orgmode export to pdf needs <WesterWest[m]>can i call guix system reconfigure / guix home reconfigure from guile? <WesterWest[m]>I want to create an install script for my configuration and makefile isn’t that great as I need to pass it multiple arguments <bjc>i don't know specifics, but the guix command is, itself, a scheme file <bjc>so there's no reason you can't use its machinery to do what you want <grobx[m]>I'm trying to compile something using ccl in a `guix shell ccl openssl@1.1.l`; but it tells me that it cannot find any suitable libssl.so.*; if I export LD_LIBRARY_PATH to some /gnu/store/*openssl, then it works fine. How can I set LD_LIBRARY_PATH according to the correct openssl path of this guix shell? <tissevert>dunno, where does that variable come from ? I see it manually set during the build phases of several packages but it seem to be related to no language in particular <grobx[m]>i don't know. Can you point me to some package that set LD_LIBRARY_PATH? <grobx[m]>I just found a /gnu/store/*openssl path that contains a libssl.so.1.1 and export LD_LIBRARY_PATH to that path; I'm wondering how to find the correct /gnu/store/*openssl related to the current guix shell <grobx[m]>I think I found a way using `export LD_LIBRARY_PATH=${GUIX_ENVIRONMENT}/lib` <tissevert>indeed, this is the way to make sure you're using the one containing the openssl version you've added to the environment <tissevert>but I'm kinda curious why this isn't done automatically if that's required <tissevert>is that a runtime or compile-time error you're getting about libssl.so ? <grobx[m]>compile time, I'm trying to compile acl2 from git repository (version 8.4) <tissevert>so you're doing pretty much what the other packages are doing, so it doesn't look that surprising that you have to <tissevert>are you trying to package acl2, while you're at it ? <grobx[m]>I dont know enough of guix for trying this, so I'm just trying to setup a reproducible shell in which I can use acl2 (make basic for now) <tissevert>that's pretty much what the packages I've listed are doing too : ) <grobx[m]>`guix shell ccl openssl@1.1.1l` is the first step, now I would like to make a manifest.scm and try to setup a proper shell by running `eval $(guix shell --search-paths --check)` <tissevert>do you have a local clone of guix so that the grep listing I've sent can be of any help and you can look at the files I found ? <tissevert>I'm not sure but I don't think you can set variables from a manifest <grobx[m]>yes I have a local clone of guix, I'm looking at your grep list right now <grobx[m]>there should be a way, in my manifest I have emacs-next-pgtk; this export EMACSLOADPATH, so there are chances that it can be done somehow; of course I'm not sure, ill try to dig into it <tissevert>I hope something looks close enough to what you're doing <tissevert>yeah, I suppose a package could export it but I didn't find any that does after quickly looking at that list and that what made me think that perhaps it was a more complicated situation than the usual PYTHONPATH stuff <meso>my computer went to sleep, unsure if anyone helped with my question about my emacs daemon herd service. <meso>reposting: hello, I am trying to setup a emacs daemon shepherd service and was wondering if anyone can check it out to see if i am doing it correctly. https://paste.debian.net/1238167/ the issue I am having is that the service never shows its running with herd status, and when I do start it, it runs like 6 times when it was already succesful the first time it was executed. <meso>tissevert: thank you for that, seems not. sorry for reposting. <WesterWest[m]>must all packages be fully bootstrappable? ie. is dotnet never happening? <maximed>vagrantc: You could give diffoscope a LIBEXEC_PATH search path ***zmv is now known as notzmv
<bricewge>PSA: when using `guix deploy` if you get an error such as “"channel-open-session" "Channel opening failure: channel”, you need to increase sshd's “MaxSessions” value <bricewge>The default value is 10, which can be reach by `guix deploy` in some case <bricewge>`guix` need to be patched to close those sessions ***AryaK[m] is now known as Aryaaryakml[m]
***Aryaaryakml[m] is now known as Arya
***Arya is now known as AryaKiran
<WesterWest[m]>i am somewhat curious, what is the purpose of ~/.guix-home/files? I have tried to deploy my setup to a new computer, but the fish configuration file, which should be in ~/.config/fish/config.fish is missing <WesterWest[m]>I can however locate that file in .guix-home/files/config/fish/ <lilyp>WesterWest[m]: you're missing a dot in "config" <WesterWest[m]>lilyp: i know, but I dont specify the location I simply use home-fish-service-type, without any modification <bovid-19>Turn out the EFI partition wasn't mounted. `reconfigure' worked fine once I mounted it. <WesterWest[m]><WesterWest[m]> "well i think i found that it’s a..." <- should I report it or just submit a patch? <jgart[m]>lilyp: jpoiret: WDYT of the newt installer being able to configure services explicitly? <jgart[m]>for example, to "graphically" walk a user through configuring all the details of the mumble service <jgart[m]>and out comes a ready to go mumble server config <jgart[m]>we could do the same for setting up a cgit and ask the user for an initial list of git repository clone urls <vagrantc>i could see a services configurator being useful outside of the context of the installer too <jgart[m]>out comes a guix system cgit config pre-populated with the code for cloning the initial list of git repos that the user would like to share on their cgit <jgart[m]>for an independent Guix System VPS service provider <jgart[m]>Web UI walks the user through setting up a cgit, mumble, etc... <jgart[m]>then proceeds to deploy it on a rented VPS <jgart[m]>Out comes a emacs config, maybe managed by guix-home or vanilla elisp files with accompanying manifest for setting up the emacs environment ***iyzsong- is now known as iyzsong
<morganw>Does anyone know if it is possible to manage files in /etc/guix/ ? I tried with etc-service-type but trying to create a file returned an error that the directory already exists. <lilyp>morganw: the acl gets instantiated through guix-service-type, the rest I'm not sure <morganw>I was trying to set a system-wide channels file as a default, but I think the same issue would exist to create /etc/guix/machines.scm. It seems strange that the instructions say to manually configure the files. <lilyp>you do realize that you're working towards a bootstrap nightmare, right? <morganw>It seems a preferable option to relying on a file scoped to the user profile, or remembering to manually specify the channels on every pull. <civodul>morganw: yeah it'd be nice if we could specify /etc/guix/{machines,channels}.scm in the operating-system declaration <civodul>it's possible right now by extending etc-service-type, but it's inconvenient <morganw>When I tried it seemed to fail because /etc/guix already existed. <morganw>I retried just to double check the error message, maybe using a symlink instead of writing the file contents is the problem? guix system: error: symlink: File exists: "/etc/guix" <bjc>that sounds like you're trying to write to a literal file "/etc/guix" not something in that directory <morganw>(simple-service 'guix-channels-config etc-service-type `(("guix/channels.scm" ,(local-file "channels.scm")))) <bjc>i'm not familiar with that particular service, and the source code, at first glance, is not particularly enlightening either <bjc>have you tried to strace the run to see what it's actually doing? <morganw>Are there instructions anywhere on how to use strace? <bjc>strace -fo some-file.out guix system reconfigure config.scm <bjc>some-file.out will have the syscall level details, which should show what operation is throwing that error <morganw>symlink("/etc/static/guix", "/etc/guix") = -1 EEXIST (File exists) <bjc>yeah, so it's trying to clobber /etc/guix with a symlink. couldn't tell you why it's doing that, but it definitely should not be <bjc>instead of etc-service-type, have you tried ‘local-file’? <bjc>with a simple-service-type <morganw>simple-service-type doesn't seem to exist, bjc <bjc>dang, i thought i remember that as the name <bjc>lemme see if i can find the real one <bjc>my bad, it's apparently: simple-service <civodul>bricewge: neat! was it reviewed already? <civodul>i vaguely recall discussing it, but i'm not sure <civodul>looks like a much welcome change anyway :-) <bricewge>I need to go back to all that patches I have let on the side :/ <bricewge>civodul: Since you are here, I've found a ressource leak in `guix deploy`, it doesn't close ssh session <bricewge>I think it appeared near 7f20e59a13a6acc3331e <bricewge>Setting sshd `MaxSessions 2` on the remote host should trigger the issue